Molecular and Whole Animal Responses of Grass Shrimp, Palaemonetes Pugio, Exposed to Chronic Hypoxia ⁎ Marius Brouwer A, , Nancy J

Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 www.elsevier.com/locate/jembe

Molecular and whole animal responses of grass shrimp, Palaemonetes pugio, exposed to chronic hypoxia ⁎ Marius Brouwer a, , Nancy J. Brown-Peterson a, Patrick Larkin b, Vishal Patel c, Nancy Denslow c, Steve Manning a, Theodora Hoexum Brouwer a

a Department of Coastal Sciences, The University of Southern Mississippi, 703 East Beach Dr., Ocean Springs, MS 39564, USA b EcoArray Inc., 12085 Research Dr., Alachua, Florida 32615, USA c Department of Physiological Sciences and Center for Environmental and Human Toxicology, University of Florida, PO Box 110885, Gainesville, FL 32611, USA

Received 28 July 2006; received in revised form 15 September 2006; accepted 20 October 2006

Abstract

Hypoxic conditions in estuaries are one of the major factors responsible for the declines in habitat quality. Previous studies examining effects of hypoxia on crustacea have focused on individual/population-level, physiological or molecular responses but have not considered more than one type of response in the same study. The objective of this study was to examine responses of grass shrimp, Palaemonetes pugio, to moderate (2.5 ppm DO) and severe (1.5 ppm DO) chronic hypoxia at both the molecular and organismal levels. At the molecular level we measured hypoxia-induced alterations in gene expression using custom cDNA macroarrays containing 78 clones from a hypoxia- responsive suppression subtractive hybridization cDNA library. Grass shrimp exposed to moderate hypoxia show minimal changes in gene expression. The response after short-term (3 d) exposure to severe hypoxia was up-regulation of genes involved in oxygen uptake/transport and energy production, such as hemocyanin and ATP synthases. The major response by day 7 was an increase of transcription of genes in the mitochondrial genome (16S rRNA, cytochrome b, cytochrome c oxidase I and III), and up-regulation of genes encoding proteins involved in iron metabolism. By day 14 a dramatic reversal was seen, with a significant down-regulation of both mitochondrial and Fe-metabolism genes. Validation of the macroarray results with q-PCR showed similar up- or down-regulation at multiple time points for 9 genes. At the organismal level, our studies showed condition factor of grass shrimp exposed to severe chronic hypoxia was lower than normoxic controls during the first 7 days of the experiment, but there were no differences after that time point, or in grass shrimp exposed to moderate hypoxia. Surprisingly, chronic hypoxia appeared to enhance grass shrimp reproduction; females exposed to moderate hypoxia had higher fecundities and a greater percentage produced first, second and third broods than normoxic shrimp. The hypoxic shrimp took longer to produce their first brood than the normoxic controls, although starved larvae from hypoxia-exposed mothers lived longer than normoxic control larvae. Shrimp exposed to severe hypoxia also had higher fecundity than normoxic controls, although embryos from hypoxia-exposed mothers took longer to hatch than normoxic control embryos. The gene expression and reproductive results suggest that expression levels of genes encoding proteins involved in oxygen and electron transport, energy, and iron metabolism may be useful molecular indicators of both short term (b7 d) and moderate (14 d) exposure to severe hypoxia, and that chronic hypoxia may have population-level impacts on grass shrimp. © 2006 Elsevier B.V. All rights reserved.

Keywords: Gene expression; Hemocyanin; Hypoxia; Macroarray; Mitochondrial genes; Reproduction

⁎ Corresponding author. Tel.: +1 228 872 4294; fax: +1 228 872 4204. E-mail address: [email protected] (M. Brouwer).

1. Introduction responsive genes through suppression subtractive hybridization. We concentrated on those genes coding for Chronic and intermittent/cyclic hypoxia is of in- proteins in the mitochondrial electron transport chain, creasing concern as related to declines in habitat quality ATP synthesis, oxygen transport, carbohydrate metab- in coastal and estuarine environments (Diaz and Rosen- olism, protein synthesis/repair/degradation, antioxidant berg, 1995; Buzzelli et al., 2002). Hypoxia can lead to defense and lipid metabolism that are known to be rapid as well as long-term cellular, physiological and responsive to hypoxic stress (Hochachka et al., 1996; behavioral changes in a variety of organisms. Because Czyzyk-Krzeska, 1997; Hochachka and Lutz, 2001). of this, detection of short-term “rescue” responses and We found that the selected genes were significantly up- long-term adaptive adjustments caused by hypoxic ex- regulated or down-regulated when grass shrimp were posure is important in environmental research. Labo- exposed to moderate or severe chronic hypoxia. Further- ratory experiments have shown that, when possible, fish more, gene expression varied with duration and severity and crustaceans will avoid or move out of hypoxic of dissolved oxygen exposure, and hypoxia exposure conditions (Wannamaker and Rice, 2000; Wu et al., resulted in marked effects on shrimp egg production and 2002). Physiologically, aquatic invertebrates respond to larval survival. hypoxia by regulating oxygen transport by increased cardiac output and hemoglobin/hemocyanin synthesis 2. Materials and methods and expression (Mangum, 1997; Terwilliger, 1998; Paul et al., 2004). At the molecular level, differential gene 2.1. Experimental animals, exposure methods and expression in fishes reflects the metabolic roles of tis- reproduction sues during hypoxia exposure (Gracey et al., 2001; Ton et al., 2002, 2003; van der Meer et al., 2005). Hypoxia- Grass shrimp were collected in the vicinity of Ocean responsive genes and proteins have recently been iden- Springs, Mississippi in Davis Bayou using dip nets. tified in blue crab, Callinectes sapidus (Brown-Peterson Adult females and males were segregated by sex based et al., 2005), suggesting molecular indicators show on morphological differences in the first and second promise for identifying signs of hypoxia exposure in pleopods (Meehean, 1936) and maintained in the labo- estuarine crustacea. However, these molecular signals ratory at 15 psu and 27±1 °C for 7 to 30 d prior to by themselves do not provide information on the effects experimentation. During acclimation, shrimp were held of hypoxia on the individual and its ability to help in 296 L tanks with static renewal of seawater. During maintain the population. Changes in reproductive para- acclimation and experimentation periods, grass shrimp meters in response to hypoxia have population level were fed brine shrimp nauplii once daily and commer- consequences (Wu, 2002), yet little is known about the cial flake food once daily. During all acclimation and effects of hypoxia on reproductive fitness in estuarine experimentation periods, shrimp were held in artificial organisms routinely exposed to low oxygen conditions. seawater (Fritz Super Salt, Fritz Industries, Mesquite Therefore, the aim of our studies was to link molecular TX) diluted to 15 psu with non-chlorinated well water. indicators to reproductive endpoints. Understanding the Four separate laboratory experiments were conduc- potential relationship between molecular and organis- ted to determine the effects of moderate (2.5 ppm dis- mal endpoints could reveal new mechanisms of hypoxia solved oxygen, DO) or severe (1.5 ppm DO) chronic tolerance/adaptation and may help predict ecologically hypoxia on gene expression and reproduction in grass relevant consequences of hypoxia. shrimp. The exposures were conducted in a modified We used the hypoxia tolerant, estuarine grass shrimp, intermittent flow-through system previously described Palaemonetes pugio, to examine the effects of chronic (Manning et al., 1999). The flow through test system hypoxia on gene expression and reproduction. This provided 1 L every 20 min (resulting in 3 complete species has been shown to be uniquely physiologically volume additions/day) to each of the 35 L test aquaria adapted to stressful tidal marsh habitats (Welsh, 1975). using a separate water delivery partitioner for each of the The use of a commonly occurring resident species in normoxic and hypoxic treatments. Oxygen levels were these studies allows laboratory results to be more easily controlled by bubbling nitrogen into a holding tank related and applied to field measurements. In contrast to which gravity fed to the partitioner used to deliver flow- previous studies which examined global responses of through hypoxic seawater. A 24 h timer was used to hypoxia on gene expression (Gracey et al., 2001; Ton activate a solenoid valve which controlled nitrogen et al., 2002, 2003; van der Meer et al., 2005), we have introduction into the holding tank at intervals that taken a directed approach to identify potentially hypoxia maintained oxygen in the holding tank at a level which 18 M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 resulted in the desired oxygen concentration when analyzed for gene expression due to their small size and introduced into the test aquaria. An additional partitioner insufficient tissue for analysis. Shrimp were anesthetized provided flow-through normoxic seawater, and normoxic in ice water, and the total length (TL, mm) and egg-free wet conditions were maintained by gently bubbling oxygen weight (ww, 0.1 mg) were recorded for each shrimp. The into the cells of the water partitioner prior to delivery of thorax was removed and stored at −20 °C in RNA later water to the individual aquaria. In all experiments, (Ambion, Inc. Austin, TX) for gene expression analysis. oxygen was monitored continuously in one hypoxic flow-through aquaria, and DO, temperature and salinity 2.3. Experiments 1 and 2; reproductive sampling were measured in all flow-through aquaria once or twice daily using a YSI Model 600XLM data sonde. At the end of two weeks under hypoxic or normoxic Grass shrimp were housed individually or in repro- conditions, males and females from the moderate (ex- ductive pairs in retention chambers constructed from periment 1) and one severe (experiment 2) hypoxia 10 cm Petri dish bottoms with a 15 cm high collar of exposure were paired into reproductive groups to deter- 500 μm nylon mesh and a 10 cm diameter disposable mine differences in fecundity and survival of the F1 Petri dish lid to prevent escape. The mesh walls of the generation based on hypoxia exposure. Hypoxic males chamber facilitate flow of water into the chamber. were paired with hypoxic females in 16 individual Twenty-five chambers were placed into each of the 35 L breeding chambers under continued hypoxic conditions flow-through glass aquaria in a water bath held at 27± (moderate or severe). All other mating pairs (hypoxic 1 °C. The exchange of water within the chambers was males (16)×hypoxic females (16); hypoxic males (16)× assured by fluctuating the water level within the aquaria normoxic females (16); normoxic males (16)×hypoxic 8 to 10 cm (18.6 to 23.3 L) periodically with a self- females (16) and normoxic males (16)×normoxic starting siphon. This compartmentalization of the test females (16)) were kept under normoxic DO. Pairs organisms precluded cannibalism and enabled individ- were checked daily for egg production, and sacrificed ual identification and the enumeration of molts of each after the female was determined gravid for a minimum test animal. Retention chambers were maintained at a of 2 days. All eggs from each sacrificed female were minimum depth of approximately 7±1 cm and a maxi- removed and counted, and 20 viable eggs from each mum depth of 10±1 cm. female were incubated individually in sterile seawater in 24-well polystyrene culture plates in a stirring incubator 2.2. Experiments 1, 2 and 3; sampling for gene expression at 27 °C and 60 rpm for 12 d. Culture plates containing embryos were observed daily and percent embryo One moderate (experiment 1) and two severe (expe- survival was determined by successful hatch by day riments 2 and 3) hypoxia studies were conducted follo- 10 post-isolation. The reproductive portion of the expe- wing similar protocols to monitor gene expression as riments lasted for 4 weeks. Sacrificed females from the well as reproductive effects of hypoxia. In all experi- reproductive portion of the study were processed for ments, 310 shrimp were isolated individually into 12 gene expression analysis as described above. At the aquaria. Male (2 tanks×25 shrimp) or female (6 tanks× termination of the experiment, all remaining females 25 shrimp) shrimp were maintained in 8 hypoxic aquaria that had not produced egg masses were sacrificed and (2–3 ppm DO, moderate hypoxia or 1.5 ppm DO, severe processed for gene expression analysis. hypoxia). The normoxic shrimp (6–8 ppm DO) were housed in 1 tank of 35 males and 3 tanks of 25 females. 2.4. Experiment 4; multiple brood reproduction For each experiment, twenty female shrimp (10 normoxic, 10 hypoxic) were sampled at three time points A fourth experiment to determine the effects of mo- (days 3, 7, and 14) during the course of the studies for derate hypoxia on reproduction was undertaken to exa- analysis of gene expression, for a total of 60 individual mine the production and condition of multiple broods shrimp per experiment. An equal number of shrimp produced by the same female. For this experiment, 100 were removed from each 35 L aquarium during each male: female pairs of shrimp were isolated into each of 4 sampling event to maintain similar densities of shrimp in moderate hypoxic (2–3 ppm DO) aquaria and 4 nor- each aquarium. At the beginning of each experiment, 10 moxic (6–8 ppm DO) aquaria for 10 weeks, resulting in female shrimp were removed from the test population 25 reproductive pairs/aquaria. During the study, females for the day 0 assessment. Sample selection was made with egg masses were isolated into hatching chambers from females at each sampling time that had egg masses, within their aquaria (5 cm petri dish with a 10 cm collar and eggs were removed and counted. Males were not of 1 mm mesh) 7 to 8 d after first observation of the M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 19

Table 1 Genes used for construction of macroarrays to detect hypoxia-responsive genes in Palaemonetes pugio Protein folding/repair and translocation 1 Peptidyl prolyl cis–trans isomerase P54985 a, b 5.81 e−24 c 2 Hsp70 cognate BAA32395 a, b 2.80 e−45 c 3 Mitochondrial import inner membrane translocase subunit (TIM14) NP648475 a, b 1.94 e−27 c 4 Hsp70 AAL27404 a, b 3.00 e−63 c 5 Hsp70 (AY935982 d) Z15041 a, b 3.00 e−75 c 6 Signal sequence receptor-translocation associated protein AAK15544 a 2.40 e−31 c

Protein synthesis 7 Ribosomal protein L13 Q90Z10 a 1.26 e−34 c 8 Ribosomal protein L5 Q26481 a 7.73 e−19 c 9 Ribosomal protein S6 AY769320 a, b 1.00 e−57 c 10 Ribosomal protein L31 Q91A76 a 3.09 e−29 c 11 Ribosomal protein L27A BAC54559 a, b 2.03 e−25 c 12 Mitochondrial ribosomal protein S2 NP057118 a, b 9.66 e−42 c 13 Ribosomal protein L21 AAK95147 a 2.00 e−98 c 14 Ribosomal protein L3 AAH42242 a, b 2.00 e−78 c 15 Elongation factor 2 (EF2) AAR01298 a, b 2.00 e−74 c 16 Ribosomal protein L6 NP498584 a 1.00 e−15 c 17 Ribosomal RNA 16S (16S rRNA) AF3047 a, b 1.00 e−18 c 18 Ribosomal protein S20 (AY935983 d) NM079697 a 1.00 e−27 c 19 Ribosomal protein S14 (AY935984 d) D14609 a, b 9.00 e−16 c

Protein degradation 20 Cathepsin L BAC65418 a, b 1.0 e−96 c 21 Aminopeptidase N XP396261 a 2.0 e−29 c 22 Cathepsin C BAC57934 a 5.2 e−30 c 23 Cysteine proteinase CAA75309 a, b 9.0 e−11 c 24 Trypsin CAA75309 a, b 1.8 e−08 c 25 Crustapain (cysteine proteinase) BAC65417 a, b 3.0 e−56 c 26 Aminopeptidase N AAD09272 a 1.7 e−07 c

Lipid metabolism 27 Acetyl CoA binding protein P12026 a, b 1.5 e−22 c 28 Acetyl CoA dehydrogenase NP776919 a 1.6 e−04 c 29 Dehydrocholesterol reductase BAD51990 a 4.0 e−61 c 30 Lipase I 046107 a, b 1.6 e−18 c 31 Vitellogenin-1 BAB69831 a, b 1.0 e−108 c 32 Apolipoprotein A-1 CAC34942 a, b 9.0 e−72 c 33 Fertilization envelope AAD23572 a, b 1.6 e−05 c 34 Vitellogenin-2 AAG17936 a, b 4.0 e−05 c

ATP synthesis and electron transport 35 Cytochrome c oxidase subunit III CAB40368 a, b 5.02 e−32 c 36 Cytochrome b (cytB) NP038299 a, b 2.17 e−38 c 37 ATP synthase d chain (ATPsyn-d) Q24251 a, b 9.96 e−23 c 38 Cytcochrome c oxidase subunit III (Ccox III) CAB40368 a, b 2.47 e−26 c 39 ATP synthase f chain (ATPsyn-f) Q9W141 a, b 4.58 e−41 c 40 ATP synthase b chain AAH61296 a, b 3.00 e−08 c 41 Cytochrome c oxidase subunit I (Ccox I) CAG26687 a, b 2.00 e−97 c

Oxygen transport and sensing 42 Hypoxia inducible factor 1a (AY655698 d) XP967427.1 a, b 4.0 e−122 c 43 Haemocyanin 1 (HcyI) AAF04148 a 1.0 e−129 c 44 HEMOCYANIN (HcyII) P80888 a, b 8.2 e−32 c 45 Hemocyanin (HcyIII) AAL27460 a, b 3.0 e−90 c 46 Hemocyanin subunit 4 (HcyIV) CAD56697 a, b 1.0 e−47 c (continued on next page) 20 M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31

Table 1 (continued ) Carbohydrate metabolism 47 Drosophila melanogaster CG1637-PA, isoform A NP727464 a 8.8 e−04 c 48 Amylase I CAB65552 a, b 1.0 e−129 c 49 Alpha-amylase preproprotein AAO72321 a 2.0 e−22 c 50 Acid beta glucosidase P17439 a, b 2.4 e−31 c 51 PEP carboxykinase CAB85964 a, b 1.0 e−108 c 52 Chitinase AAN74647 a, b 2.1 e−21 c 53 Glycogenin-1 P13280 a 2.6 e−12 c

Cell structure/motility and muscle contraction 54 Troponin C gamma NP001011651 a, b 3.0 e−60 c 55 Alpha-1-tubulin AAC47522 a 6.0 e−87 c 56 Cellular myosin A71144 a 0 c 57 Fast myosin heavy chain AAA17371 a 3.0 e−40 c 58 Troponin I, fast skeletal muscle P05547 a 1.0 e−13 c 59 Beta-actin (AY935989 d) AY626840 a, b 4.0 e−139 c

Metal binding and anti-oxidant 60 Heme binding protein NP956492.1 a, b 4.0 e−11 c 61 Ferritin subunit XP624076 a, b 4.0 e−9 c 62 Cytosolic Mn-superoxide dismutase (cyt-MnSOD) (AY211084 d) DQ073104 a, b 3.0 e−88 c 63 Mitochondrial Mn-superoxide dismutase AE017283 a, b 4.0 e−142 c (mit-MnSOD) (AY935986 d) 64 Cd metallothionein 1 (CdMT1) (AY935987 d) AAB5227.1 a 5.0 e−18 c

Blood Coagulation and Immune function 65 PmAV S78774 a, b 3.9 e−05 c 66 Clottable protein AAF19002 a 3.4 e−37 c 67 Beta-1,3-glucan binding protein AAM21213 a, b 2.0 e−51 c 68 Coagulation factor V and VIII CAC94896 a 2.0 e−17 c 69 Complement C3-S BAA36621 a 2.0 e−55 c 70 Fibrinogen A AAH41754 a 1.2 e−41 c

Miscellaneous functions 71 Cutical protein AMP4 P81388 a 3.99 e−10 c 72 Ornithine decarboxylase antizyme P70112 a 2.10 e−12 c 73 H3 histone, family 3B XP235304 a 8.60 e−44 c 74 Fibrillarin NP523817 a 3.00 e−66 c 75 Orn decarboxylase antizyme P55814 a, b 1.80 e−06 c 76 Glutamine repeat protein-1 NP032158 a, b 3.80 e−07 c Genes are arranged by biological function. a Closest match as identified by BLASTX search. b Significantly up-regulated or down-regulated based on macroarray analysis with all normalization techniques. c e(xpect) values (Karlin and Altschul, 1990). d GenBank nucleotide accession numbers: grass shrimp sequences. eggs. The hatching chamber was placed in a larger normoxic conditions and monitored twice daily until retention chamber of smaller mesh (200 μm) to retain death from starvation. Larvae were not fed during hatched larvae and separate them from the female when this time to determine whether yolk content may have the offspring hatched. Following hatch, the female was changed with subsequent broods and to determine if returned to the original retention chamber with her mate. survival differed between normoxic and hypoxic broods. This method allowed for assessment of multiple broods from each reproductive pair. All hatched larvae were 2.5. Cloning, subtractive hybridization and macroarrays counted, and approximately twenty-five larvae from of grass shrimp each female were isolated into a 200 μm mesh chamber for survival assessment. Normoxic and hypoxic larvae Eight genes (Table 1, superscript c), including heat from each female and each brood were kept under shock protein (Hsp70), mitochondrial and cytosolic M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 21 manganese superoxide dismutase (mit-MnSOD and cyt- determined for each gene. Quality of macroarrays was MnSOD), Cd metallothionein (CdMt1), hypoxia induc- further evaluated by determining correlation coefficients ible factor 1α (hif-1α), β-actin and ribosomal proteins and slopes of scatter plots of duplicate signal intensities S14 and S20 were cloned and sequenced from grass for all macroarray combinations. shrimp thorax tissue containing hepatopancreas using RT-PCR. An additional 68 potentially hypoxia respon- 2.6. Real-time PCR sive genes (Table 1, superscript a) were identified from two suppression subtractive hybridizations (SSH) per- To validate results from the gene arrays, real-time formed by EcoArray Inc, Alachua, FL. Subtracted quantitative RT-PCR (q-PCR) was run for genes that libraries were constructed with poly-A mRNA, con- showed significant up-or down-regulation on the macro- verted to cDNA, that was isolated from day 0 normoxic arrays. These included hemocyanin (3 separate subunits), (control) and day 3 and 5 moderate hypoxic grass ATP synthase f chain, cytochrome c oxidase (I and III), shrimp and with mRNA from day 0 controls and day 3 cytochrome b, 16S rRNA and ferritin. The q-PCR was severe hypoxic shrimp. Subtractive hybridizations were performed in both directions on these samples in order to Table 2 obtain up-regulated and down-regulated genes. Subtrac- Sequences and percent amplification efficiency of primers used for tive hybridizations were performed using the Clontech real-time PCR validation of macroarrays (Palo Alto, CA) SSH kit following the manufacturer's Gene Forward primer-5′ Reverse primer-5′ Efficiency recommendations. The resultant pool of cDNA clones Haemocyanin 1 AAA CCA GTA AGG ACG GGC 94% 92% were shotgun ligated into pGEM T-Easy cloning vector (HcyI) GAA GAG GGC AGG GAA TTCa (Promega, Madison, WI), transformed into DH5α cells, TTT GCTa and and TGT CAC AGG GCT TTG GAA CTC TTG and plated onto Luria-Bertani (LB) agar plates contain- b μ CTC CAC ACA CTA CTC C ing ampicillin and oxacillin (100 g/mL each). Recom- Cb binant colonies were picked from the plates, plasmids ATP Synthase f ATG GAC CGA AAT TTC CTT 94% were purified and inserts were sequenced. (ATPsyn-f) GAC TTT CCC CTC CAC CCT 101% The resulting 76 genes produced from cloning and AGA Ta and CGC CCT Ta and AAC SSH were PCR amplified and then robotically spotted in CGA AGC ATT TGC CCA TAA GGA GGT Gb GGA GCG AAT duplicate onto neutral nylon membrane macroarrays Cb together with various controls, including exogenous α-tubulin AGA CTG TGC CGA TGA GAC 94% 94% Arabidopsis “spiking” genes, as previously described CTT CAT GGT GGT TCA AGT (Larkin et al., 2003). Genes were arranged on the mem- CGA Ta and GCA TGG Ta and AGT GGT CCG CTG TGG TGT AGG brane by functional group (Table 1). Total hepatopan- b b – TTG TTG TGG GTC TC creatic RNA was extracted from 8 10 grass shrimp per HEMOCYANIN GCA GTC ACT GGA TAC GGC 98% treatment group using Stat-60 (Tel-Test, Friendswood, (HcyII) GAT GGA GAA GAT CAA GAG TX). Genomic DNA was removed by DNase treatment TAT Gb Gb and total RNA was transcribed into radiolabeled cDNA Hemocyanin ATA CAG TCC AAA CCA GGG 92% b and hybridized to the membranes. Background subtrac- (HcyIII) CAA TGC TCA TCA CGA GTG ATA Cb tion for each cDNA spot was performed as previously Cytochrome CGG AGC GTG CGA AGG CGT 94% described by Larkin et al. (2003). The values were then c oxidase I AGC AGG AAT GGG CTG TAA b b log2 transformed and normalized three different ways (CcoxI) AG C (to the mean and median intensity of the array data, as Cytochrome TTG TTG CCA CCA GGC TGC 98% well as α-tubulin). Fold change values, calculated from c oxidase III CAG GAT TCC TGC TTC AAA (CcoxIII) ATGb Gb the mean values of normalized hypoxic and normoxic Cytochrome ACC CTT TTA CCT TAT TTG 91% shrimp for each time point were used to determine which b (CytB) ACG CCA TAC TCT GGG ATA gene transcripts were up-regulated or down-regulated by ATA Cb GAG Cb hypoxia. 16S rRNA TTG TAA GGG TAC GCT GTT 96% To assess consistency of gene signal intensities on the TAG CTG TGT ATC CCT AAA Gb Gb macroarrays, a pooled sample of extracted RNA was Ferritin CGT GGT GGA GGT TCA TAC 103% reverse transcribed into labeled cDNA and hybridized to AGC ATC AAT TCT TGG TCA b b 4 separate membranes. Values were log2 transformed G CTT C and normalized to the median array intensity, and the Primers designed using aPrimer Express Software or bBIO-RAD coefficient of variation among the 4 membranes was iCycler iQ Beacon Designer software. 22 M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 also used to validate the utility of α-tubulin for use in used for the PCR reactions with the following cycle normalizing the array data. Primers for these genes were parameters: 1 cycle of 95 °C for 2 min, 50 cycles of designed using Primer Express software (Applied Bio- [95 °C for 15 s and 58 °C for 15 s]. systems, Foster City, CA) and BIO-RAD iCycler iQ Each sample was run in duplicate for both q-PCR Beacon Designer software. Sequences for these primers protocols using 5–10 hypoxic or normoxic shrimp. The are shown in Table 2. averaged duplicate Ct value (PCR cycle threshold where Each primer set was validated for specificity and target amplification is first detected) was normalized to efficiency by running dissociation and standard curves, measured 18S rRNA Ct values for each sample. 18S respectively. Standard curves included data from a rRNA values did not fluctuate between treatment groups −ΔΔCt minimum of four serially diluted cDNA samples. The (data not shown). The comparative Ct (2 ) method efficiency of amplification for each primer set is shown of analysis was used to determine changes in gene in Table 2. The amplification efficiency of 18S rRNA, expression between controls and treated samples (Wong which was used as a normalizing gene, was 94%. and Medrano, 2005). Different protocols were used for q-PCR validation of the moderate and severe hypoxia treatments. For the 2.7. Data analysis moderate hypoxia samples, DNase-treated (DNA-free; Ambion, Inc., Austin, TX) total RNA from hypoxic and Throughout all experiments, each shrimp in its indi- control grass shrimp was reverse transcribed to cDNA vidual chamber is considered a replicate for both gene using random hexamers and Multiscript reverse tran- expression and reproduction (n=10) because maintain- scriptase according to the manufacturer's instructions ing hundreds of shrimp in individual experimental units (Applied Biosystems). The q-PCR was performed in at the same DO is practically impossible. Additionally, 25 μl reactions that contained 100 ng shrimp cDNA, the flow-through system and high volume of water 12.5 μl SYBR green master mix (Applied Biosystems exchange in each experimental tank limited interaction P/N 4309155; which included SYBR green, buffer, Taq through chemical cues among shrimp in the same tank, polymerase, and dNTPS), and 50 nM each of the for- whereas the separation into individual retention cham- ward and reverse primers. The PCRs using 18S rRNA bers prevented physical interactions. primers were also performed in a 25 μl reaction and A length-weight relationship was calculated for all contained 0.5 ng cDNA, 12.5 μl SYBR green master grass shrimp sampled (ww=14.84+3.741TL, r2 =0.863, mix, and 50 nM each of the 18S rRNA forward and pb0.001). The slope of this line was used to calculate reverse primers (Applied Biosystems). For all of the condition factor (K) for each individual (K=ww/TL3.74 × samples, minus RT controls were run to ensure the 100,000; see Murphy and Willis, 1996). There was a removal of all contaminating genomic DNA. An Applied significant relationship between shrimp length and num- Biosystems 7500 thermocycler was used for the PCR ber of eggs (TL=8.39 eggnum — 90.63, r2 =0.33, reactions with the following cycle parameters: 1 cycle of p=0.002). Therefore, the relative fecundity of grass 50 °C for 2 min, 1 cycle of 95 °C for 10 min and 40 cycles shrimp was calculated by dividing the number of eggs or of [95 °C for 15 s and 60 °C for 60 s]. hatched larvae by the ww of the female (expressed as For validation of the severe hypoxia samples, DNase- # eggs/g), and is used in all analyses. Reproductive groups treated (DNA-free, Ambion,) 1 μg total RNA from from the moderate and severe hypoxia studies were hypoxic and control grass shrimp was reverse tran- combined into 3 groups, based on the DO history of the scribed to cDNA using random hexamers (Ambion) and females; females continuously exposed to hypoxia (HH), Superscript II Reverse Transcriptase according to the females continuously exposed to normoxia (NN) and manufacturer's instructions (Invitrogen). The cDNA females exposed to hypoxia the first 2 weeks of the study was diluted 40× with sterile water for q-PCR. q-PCR but allowed to mate in normoxia (HN). Differences in was performed in 50 μl reactions that contained 25 μliQ relative fecundity, K, percentage of hatched embryos, and SYBR Green Supermix, (BioRad, cat # 170-8884, days to hatch among the 3 treatment groups for the mo- which included SYBR Green 1, buffer, iTaq DNA derate and severe hypoxia studies were tested using polymerase and d-NTP's), 400 nM each of the forward ANOVA; differences among groups were evaluated with and reverse primers and 2 μl of the diluted cDNA. the Bonferroni post-hoc test. Differences in K, larval Minus RT controls were run for some samples to ensure survival, interbrood interval, and larval survival between the removal of all contaminating DNA, and in all cases normoxic and moderate hypoxic females in the multiple no contaminating DNAwas found. A BIO-RAD I Cycler brood study were tested using student's t-test. Percentage iQ Multi-Color Real Time PCR Detection System was data were arcsine square root transformed prior to analysis M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 23

(Sokal and Rohlf, 1995). Data were tested for homoge- molting frequency between normoxic and hypoxic neity of variance (Levene's test) and normality of shrimp were not quantified since shrimp held in nor- distribution (1 sample Kolmogorov–Smirnoff test) and moxia rapidly consumed their molts (often in b24 h) and were log-transformed if necessary to meet these assump- we feel that molting frequency in normoxic shrimp is tions. Student's t-test and ANOVA were performed with thus underrepresented in our observations. It is note- SPSS (version 11.5). Data were considered significant if worthy, however, that shrimp held in both moderate and p≤0.05. severe hypoxia rarely consumed their molts, which Gene expression data from macroarrays normalized to often persisted in the cages for 2–3d. the array mean, median, and α-tubulin, and q-PCR data normalized to 18S rRNA were analyzed with Stu- 3.2. Gene expression dent's t-test to determine differences between normoxic and hypoxic grass shrimp for each time point in the Forty-eight of the original 76 potentially hypoxia- moderate and severe hypoxia experiments. Only shrimp responsive genes showed significant up-or down-regula- continuously exposed to normoxia or moderate/severe tion for all 3 normalization methods (mean, median, α- hypoxia were used for these analyses; shrimp from the HN tubulin) for at least one time point (Table 1,superscriptd). reproductive group were not included in these ana- Data presented here represent these most robust genes, lyses. Changes in gene expression for both macroarrays and are displayed using α-tubulin normalization, since and q-PCR were considered significant if p≤0.05. The this gene was determined to be a consistent normalization gene expression data are plotted as fold change, and thus gene (see real-time PCR section below). error bars for the control and treated samples are not shown. Tests of variability of gene signal intensities on the macroarrays due to experimental error showed a coef- 3. Results ficient of variation ≤20% for the robust genes. Com- parisons among the membranes showed slopes of scatter 3.1. Grass shrimp survival and condition plots of signal intensities of corresponding genes for all six macroarray combinations ranging from 0.835 to Survival of grass shrimp during the moderate hypoxia experiment was excellent; only 1 out of 225 females died during the course of the 46 d study. Mortality during the 14 d exposure period of severe hypoxia was 2.7% for both hypoxic and normoxic females. During the 4.5 week reproductive portion of the severe hypoxia study, female mortality was 22.6% in hypoxia and 10.3% in normoxia; no males died during the course of this study. Female mortality was higher during the 10 week multiple brood moderate hypoxia experiment, 32% in normoxia and 15% in hypoxia. There were no significant differences in length or weight between normoxic and hypoxic grass shrimp during the 14 d exposure period or subsequent reproductive periods for any of the experiments. There were also no differences in condition factor between normoxic and hypoxic females for either of the moderate hypoxia studies. However, after 3 and 7 d exposure to severe hypoxia, condition factor of normoxic females was higher than that of hypoxic females ( p=0.049 and 0.052, respectively). There were no additional significant differences in condition factor as the study progressed. This suggests severe, chronic hypoxia may have a short-term effect on grass Fig. 1. Fold changes in gene regulation measured by macroarrays in shrimp condition, but they are able to adjust and com- grass shrimp exposed to chronic, moderate (2.5 ppm DO) hypoxia. Data shown are normalized to a-tubulin and all changes are significant pensate within 1 week of exposure. (t-test, p≤0.05). Genes are pattern coded by functional group, as Shrimp in both normoxic and hypoxic exposures defined in Table 1. Genes with values b−1 are down-regulated 2-fold routinely molted every 5–8 d. However, differences in or greater. A. 7 day exposure. C. 14 day exposure. 24 M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31

0.955 and r2 values from 0.83 to 0.95 indicating good significant changes in expression of a number of genes consistency of signal intensity values among membranes. that are potentially robust indicators of hypoxia. Fur- In addition, a plot of mean intensity for each gene from the thermore, gene expression profiles change over the normoxic controls (n=23) of the moderate hypoxia expo- time-course of chronic DO exposure, lending further sure experiment against the mean intensity of corres- insight into how the grass shrimp adapt to severe hy- ponding genes from the controls of the severe hypoxia poxia. After 3 d exposure to severe hypoxia, signifi- experiment (n=33) had a slope of 0.891 and r2 value of cantly up-regulated genes include ATP synthase d and f 0.792, indicating good consistency of signal intensity chains (ATPsyn-d and ATPsyn-f ), 3 hemocyanin genes values among shrimp. (Hcy II, Hcy III, and Hcy IV), troponin C and ferritin There were no significant changes in gene expression (Fig. 2A), suggesting an attempt to increase oxygen after 3 d exposure to moderate, chronic hypoxia. How- uptake/transport (hemocyanin), ATP synthesis (ATP ever, after 7 d exposure, there was significant down- synthases) and locomotion (troponin C). After 7 d ex- regulation of 2 HSP70 genes (Fig. 1A). After 14 d exposure posure to severe chronic hypoxia, the adaptation in- to moderate, chronic DO, there was a significant 19-fold duced by day 3 becomes insufficient, and ATP synthase, decrease in expression of the gene encoding the anti- hemocyanin and troponin are no longer up-regulated oxidant enzyme cytosolic Mn Superoxide Dismutase (Fig. 2B). The major response by day 7 appears to be an (cSOD), which in crustacea has replaced the more com- increase of transcription of genes present in the mito- monly found cytosolic Cu, Zn Superoxide Dismutase chondrial genome (16S mitochondrial rRNA (16S rRNA) (Brouwer et al., 2003, 1997). Expression of one other and cytochrome c oxidase subunit 1 (Ccox I); Fig. 2B). gene, which shows weak sequence similarity (E value= Increased synthesis of cytochromes, which are Fe/heme 3.83−07) with a mouse gene encoding for glutamine repeat proteins, is accompanied by up-regulation of the genes protein, is up-regulated (Fig. 1B). It appears in general that encoding heme binding protein and ferritin (Fig. 2B). The no genes are robust indicators of moderate chronic DO adaptation seen after 7 d once again becomes insufficient exposure, with the possible exception of cSOD. by Day 14, and a dramatic reversal is seen, with a signi- In contrast to moderate hypoxia, grass shrimp expo- ficant down-regulation of transcription of genes in the sed to severe (1.5 ppm DO), chronic hypoxia showed mitochondrial genome (16SrRNA, Ccox I, cytochrome c

Fig. 2. Fold changes in gene regulation measured by macroarrays in grass shrimp exposed to chronic, severe (1.5 ppm DO) hypoxia. Data shown are normalized to α-tubulin and all changes are significant (t-test, pb0.05). Genes are pattern coded by functional group as defined in Table 1. Genes with values N1 are up-regulated 2-fold or greater. Genes with values b−1 are down-regulated 2-fold or greater. A. 3 day exposure. B. 7 day exposure. C. 14 day exposure. D. 26–61 day exposure. M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 25 oxidase III (Ccox III) and cytochrome b (cytB))aswellas Moderate hypoxia resulted in few significant changes ferritin (Fig. 2C). Finally, PmAV, a novel gene shown to be in gene expression in macroarrays (Fig. 1), and q-PCR important in virus resistance in the shrimp Penaeus also showed no significant changes in expression of α- monodon, is also down-regulated after 14 d exposure to tubulin, Hcy I or ATPsyn-f, although the direction of severe hypoxia, suggesting a potential for increased sus- gene regulation (up or down) was the same for these ceptibility to disease with chronic severe hypoxia expo- genes in the macroarrays and q-PCR (Fig. 3A). sure. Prolonged 26–61 d exposure to severe hypoxia In contrast to results for moderate hypoxia, the q-PCR shows continued down-regulation of mitochondrial pro- showed a greater, but still not significant, up-regulation of teins Ccox III and cytB (Fig. 2D). Thus, mitochondrial 3 separate hemocyanin genes after 3 d exposure to severe genes such as 16S rRNA, cytB, Ccox I and Ccox III as hypoxia (Fig. 3B) mirroring macroarray results for the well as hemocyanin and Fe-proteins appear to provide same time point (Fig. 2A). Additionally, ATPsyn-f was promise as indicators of chronic severe hypoxia exposure also up-regulated in both q-PCR (Fig. 3B) and on macro- in grass shrimp. arrays (Fig. 2A). Furthermore, 2 separate groups of shrimp exposed to severe hypoxia for 3 d showed a similar 3.3. Real-time PCR validation up-regulation of the 3 hemocyanin genes and ATPsyn-f measured using q-PCR and macroarray analysis, suggest- The q-PCR was used to validate the gene expression ing this is a robust, consistent response. results from the macroarrays for both the moderate and There were significant changes in gene expression as severe hypoxia experiments, as well as to verify the measured by q-PCR after both 7 and 14 d exposure to utility of using α-tubulin as a normalizing gene. Alpha- chronic severe hypoxic (Fig. 3C and D). Three genes tubulin did not change significantly in response to 14 d (Ccox I, Ccox III and ferritin) were significantly up- exposure to moderate hypoxia or to 3, 7 or 14 d expo- regulated at day 7, and two of these (Ccox I and ferritin) sure to severe hypoxia (Fig. 3). These results, in com- were also significantly up-regulated on the macroarrays bination with the consistent appearance of α-tubulin at the same time point (Figs. 2B and 3C). Gene expres- with high intensity values on all membranes, justify the sion changed significantly for 16S rRNA and ferritin use of α-tubulin for normalization of the arrays. after 14 d exposure to severe hypoxia (Fig. 3D), and

Fig. 3. Fold changes in gene regulation measured by real-time PCR in grass shrimp exposed to chronic hypoxia. Genes with values N1 are up- regulated 2-fold or greater. Genes with values b−1 are down-regulated 2-fold or greater. Striped bars indicate the direction of the fold change was the same for both real-time PCR and macroarrays. Data shown are normalized to 18S rRNA. Significant (t-test, pb0.05) real-time PCR changes indicated by ⁎. A. 14 day exposure to moderate (2.5 ppm DO) hypoxia. B. 3 day exposure to severe (1.5 ppm DO) hypoxia. C. 7 day exposure to severe hypoxia. D. 14 day exposure to severe hypoxia. 26 M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 both these genes were significantly down-regulated on the macroarrays (Fig. 2C). There was only one instance of disagreement among all genes tested using q-PCR and macroarray analysis. Ferritin was up-regulated as measured by q-PCR and down-regulated as measured by macroarrays after 14 d exposure to severe hypoxia. Overall, the similarity in response of 9 genes at multiple time points using both q-PCR and macroarrays validates the macroarray gene expression results.

3.4. Reproduction

There were no significant differences in any reproductive parameters when considering male DO exposure history. Thus, all reproductive groups were analyzed based on female DO exposure history only. A higher percentage of grass shrimp exposed to continuous moderate hypoxia (HH) produced a first brood than females exposed to NN or HN. This difference was not observed in shrimp exposed to severe hypoxia (Table 3). Furthermore, shrimp exposed to moderate hypoxia (HH) took longer to produce a brood (22.7 ±2.3 d) than those exposed to NN (16.1±1.8 d), and this difference was significant (t24 =2.24, p=0.035). However, there was no between-group difference in time to produce a brood for Fig. 4. Relative fecundity (# eggs/g shrimp weight) of grass shrimp shrimp exposed to severe hypoxia (22.6±3.1 d, hypoxic, exposed to chronic hypoxia, expressed as mean±standard error. 22,6±2.5 d, normoxic; p=0.995). There was a signif- Different letters indicate means significantly different (ANOVA, icant difference in relative fecundity among exposure pb0.05). A. Moderate (2.5 ppm DO) hypoxia. B. Severe (1.5 ppm groups for grass shrimp in the moderate hypoxia expe- DO) hypoxia. HH-females continuously exposed to hypoxia. NN- females continuously exposed to normoxia. HN-females exposed to riment (F2,35 =5.317, p=0.010; Fig. 4A). Grass shrimp hypoxia for 2 weeks and than allowed to mate in normoxia. exposed to continuous moderate hypoxia (HH) had significantly higher relative fecundities than shrimp exposed to HN ( p=0.009, Bonferroni test) and marginally (11.8–12.1 d) for grass shrimp exposed to moderate hy- higher relative fecundities than normoxic (NN) shrimp poxia. However, there was a significant difference in days ( p=0.074, Bonferroni test). There was a more pronoun- until hatch in the severe hypoxia experiment (F2,44 = ced difference for grass shrimp exposed to severe 3.317, p=0.046), with NN embryos hatching earlier hypoxia, with relative fecundity of severe HH shrimp (11.8±0.2 d) than embryos from hypoxia-exposed mo- significantly higher than both NN and HN shrimp (F2,47 = thers (12.5±0.2 d). These results suggest exposure to 3.935, p=0.027; Fig. 4B). There were no differences hypoxia results in an increase in fecundity, but does not among groups in either percentage of embryos that affect hatching success of embryos. However, these re- successfully hatched (81.5–88.5%) or days until hatch sults may only apply to the first brood produced, as shrimp were sacrificed for RNA extraction prior to production of a subsequent brood in these experiments. Table 3 Percentage of female grass shrimp producing first broods as a function of hypoxia exposure Table 4 Dissolved oxygen HH NN HN Percentage of female grass shrimp exposed to chronic moderate (N=16 pairs) (N=32 pairs) (N=32 pairs) hypoxia (2.5 mg/ml DO) over a 70 d period producing multiple broods Moderate (2.5 mg/ml) 62.5% 46.9% 34.4% Brood number Hypoxia (N=100 pairs) Normoxia (N=100 pairs) Severe (1.5 mg/ml) 68.7% 53.1% 62.5% First 60% 49% HH, NN-females exposed to hypoxia or normoxia for the entire Second 24% 12% experiment; HN-females exposed to hypoxia for 14 d prior to mating, Third 3% 1% but allowed to mate in normoxia. M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 27

Table 5 4. Discussion Reproductive parameters of female grass shrimp in a multiple-brood, chronic moderate hypoxic study Our data indicate that changes in gene expression in Parameter Hypoxia (2.5 pm DO) Normoxia (8 ppm DO) grass shrimp hold promise as molecular indicators for Brood 1 Brood 2 Brood 1 Brood 2 exposure to chronic hypoxia. Furthermore, grass shrimp (N=56) (N=19) (N=48) (N=9) reproduction appears to be affected by chronic hypoxia Relative 202±12.9 189±21.3 226±14.3 178±82.8 exposure, suggesting population-level implications of fecundity long-term hypoxia. Previous studies examining the effects (eggs/g) of hypoxia on crustacea have focused on individual/ Interbrood 24.2±1.3 a 17.6±2.0 b 21.1±1.1 13.0±1.9 b interval (d) population-level (Coiro et al., 2000; Wu et al., 2002; Bell Larval 6.3±0.11 a 6.2±0.15 a 5.9±0.09 5.3±0.47 et al., 2003a,b; Mistri, 2004; Bell and Eggleston, 2005), survival (d) physiological (DeFur et al., 1990; Tankersley and Wieber, Shrimp were exposed continuously over a 70 d period to hypoxia or 2000; McMahon, 2001)ormolecular(Mangum, 1997; normoxia. Paul et al., 2004; Brown-Peterson et al., 2005)responses a Significant difference ( p≤0.05) between treatments within a brood. but have not considered more than one type of response in b ≤ Significant difference ( p 0.05) between broods within a treatment. the same study. Thus, our data provide the opportunity to integrate various disciplines in examining the responses of grass shrimp to chronic hypoxia. In the multiple brood experiment, a greater percent- Grass shrimp appear to respond differently to moderate age of females exposed to moderate hypoxia produced vs. severe chronic hypoxia. Significant changes in gene first, second and third broods than normoxic females expression were not seen until 7 d exposure to chronic, (Table 4). Furthermore, the percentage of females moderate hypoxia. The down-regulation of expression at producing first broods was similar to the previous expe- day 7 of both the constitutive (HSP70 cognate) and the riments for HH and NN shrimp (Table 3). There was no stress-inducible (HSP70) forms of the chaperone protein significant difference in relative fecundity between nor- HSP70, important in cellular stress defense (Kultz, 2003), moxic- and hypoxic-exposed females for either the first was somewhat unexpected. Previous studies on hypoxic or second brood (Table 5); too few females produced a responses in vertebrates and invertebrates have shown up- third brood for meaningful analysis. Hypoxic females regulation (Benjamin et al., 1990; Ton et al., 2003; van der took a longer time to produce their first brood than Meer et al., 2005; Brown-Peterson et al., 2005)orno normoxic females (t100.6 =−1.938, p=0.055), similar to change (Zarate and Bradley, 2003) of heat shock protein previous results for moderate hypoxia exposure, but expression. However, both forms of HSP70 are also in- there was no significant difference for the second brood volved in folding of newly synthesized proteins (Beck- ( p=0.163; Table 5). Both normoxic and hypoxic man et al., 1990), and a decrease in protein synthesis in females produced their second brood in a significantly response to hypoxia may account for the decrease in shorter time than their first brood (t55 =3.75, p=0.003, expression of genes involved in protein folding. A similar normoxic; t73 =2.82, p=0.009, hypoxic; Table 5). There down-regulation in expression of chaperone proteins such was no difference in the percentage of embryos that as chaperonin 10 and heat shock factor binding protein 1 hatched or days to hatch between normoxic and hypoxic was observed in gills of hypoxia-exposed zebrafish (van females for either brood. Finally, starved larvae from der Meer et al., 2005). Finally, the observed down-regu- hypoxic females survived significantly longer than star- lation of MnSOD in grass shrimp is a typical cellular ved larvae from normoxic females for both the first and response to hypoxia in both vertebrates (Russel et al., second broods (t100.6 =−2.488, p=0.014, first brood; 1995) and invertebrates (Choi et al., 2000; Brown-Peter- t26 =−2.25, p=0.033, second brood; Table 5). There son et al., 2005), presumably due to reduced production of were no differences in larval survival between the first superoxide radicals under hypoxic conditions. However, and second broods of an individual female for either this down-regulation has been previously shown in the normoxic or hypoxic larvae ( pN0.24). These results mitochondrial form of MnSOD, whereas the unusual suggest that moderate hypoxia impacts the initial ability cytosolic form of MnSOD was down-regulated in grass of females to produce eggs, but that the number of eggs shrimp. produced is not affected. However, it appears that The changes observed in gene regulation in response larvae from moderate hypoxic females have a greater to severe chronic hypoxia lend insight into potential energy store and are able to survive longer without mechanisms grass shrimp may use to survive hypoxic food. conditions. There were distinct time-course related 28 M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 differences in expression of genes important in ATP levels each increase with increased oxidative capacity of synthesis, oxygen uptake/transport and the mitochon- muscle tissue (Williams, 1986). These results indicate drial electron transport chain. The initial response after that the expression of mitochondrial genes in mamma- short-term (3 d) exposure of grass shrimp to severe lian striated muscle is proportionate to their copy num- hypoxia was upregulation of genes involved in oxygen ber, suggesting that amplification of the mitochondrial uptake/transport and energy production, such as hemo- genome relative to chromosomal DNA is an important cyanin and ATP synthases. This suggests an initial feature underlying enhanced expression of mitochon- attempt to compensate for reduced availability of drial genes in highly oxidative tissue. Thus, downregu- oxygen by increasing the oxygen transport and ATP lation of these genes in hypoxic situations could be a synthesis capacity. Similar upregulation of proteins result of decreased need for oxidative metabolism and involved in oxygen transport such as myoglobin has therefore reduced copy numbers of mitochondria and also been observed in gill tissue of zebrafish exposed to mitochondrial genes. Grass shrimp may use this mecha- hypoxia (van der Meer et al., 2005). nism to cope with hypoxic stress, and save energy by After 7 d exposure to severe hypoxia, expression of reduction of mitochondrial biogenesis. hemocyanin and ATP synthase genes has returned to Finally, down-regulation of PmAV, a gene important normoxic levels. The major response by day 7 appears in virus resistance in penaeid shrimp (Luo et al., 2003), to be an increase of transcription of genes present in the suggests the grass shrimp immune system may be com- mitochondrial genome (16S mitochondrial rRNA, cyto- promised with chronic exposure to severe hypoxia. chrome c oxidase 1 (Ccox 1) and to a lesser extent Burnett and Burnett (2000) suggested hypoxia results in (pb0.1) cytochrome b), together with upregulation of a a depression of the generalized innate immune response putative heme binding protein and the iron storage in P. pugio and Penaeus vannamei based on measure- protein, ferritin. This apparent link between mitochon- ments of circulating hemocytes and survival of shrimp drial electron transport chain and proteins involved in exposed to Vibrio. A similar conclusion was reached iron metabolism is not unexpected since the mitochon- regarding hypoxia exposure in killifish (Boleza et al., drion is a dynamo of Fe metabolism, being vital not only 2001). Thus, prolonged hypoxia may have population for heme (cytochrome) biosynthesis but also for the consequences, as individuals that have already down- biogenesis of [Fe–S] clusters that are present in more regulated their aerobic metabolism also have decreased than 10 subunits of enzymes in Complex I, II and III of immune defenses, which could result in high mortality the respiratory chain (Napier et al., 2005; Taketani, of the population. 2005). Data from macroarray and microarray analysis need The adaptation observed after 7 days apparently to be interpreted cautiously (Kothapalli et al., 2002) and becomes insufficient by Day 14, and a dramatic reversal validation of observed changes with additional mea- is seen, with a significant downregulation of transcrip- surement techniques is desirable. All the genes in this tion of genes in the mitochondrial genome (16S rRNA, study were identified using SSH, and gene expression cytochrome c oxidase subunits I and III (Ccox I and III) results on the arrays mirrored the up-or down-regulation and cytochrome b), similar to results from zebrafish seen with SSH. We also used q-PCR to validate our exposed to long-term hypoxia (van der Meer et al., array data, and found that 9 genes showing differential 2005). Both ferritin ( p=0.003) and to a lesser extent the expression on macroarrays were also differentially ex- heme binding protein ( p=0.062) are down-regulated as pressed with q-PCR. The fold change values, and sig- well. nificance of these values, between macroarrays and RT- Mitochondrial genes of the grass shrimp are simul- PCR were not always directly comparable, but in all but taneously down-regulated in response to long-term one instance up-regulation or down-regulation was con- hypoxia. Similarly, mitochondrial (Ccox I and II ) and firmed. Quantitative differences between array data and nuclear encoded (Ccox IV and Vb) subunits of cyto- q-PCR results have been reported previously (Ton et al., chrome c oxidase are coordinately down-regulated in 2002, 2003; van der Meer et al., 2005; Brown-Peterson mouse and rat cell lines during hypoxia (Vijayasarathy et al., 2005), and which of the two methods is more et al., 2003). This suggests control of their transcription accurate is debatable (Allison et al., 2005). is coordinately regulated in response to hypoxia, or it Interestingly, moderate chronic hypoxia resulted in may reflect an overall decrease in mitochondrial bio- more dramatic effects on reproduction in grass shrimp genesis with a concomitant reduction in the number of than it did on regulation of our SSH identified hypoxia- mitochondrial genome copies. Mitochondrial DNA, mi- responsive genes. Additionally, there was a wider va- tochondrial ribosomal RNA and cytochrome b mRNA riety of reproductive effects related to moderate, rather M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 29 than severe, chronic hypoxia. The surprising results of ronment, where food was not limiting, there was no increased fecundity, percentage of ovigerous females predation threat, and movements were restricted to a small and larval survival of hypoxia-exposed shrimp cannot area. be readily explained by classic life history theories. Our results suggest that gene expression may be a However, it has been proposed that Daphnia channel useful indicator for measuring both short term (b7d)and more resources into growth and reproduction early in moderate (7–14 d) exposure to severe chronic hypoxia. life when faced with sub-optimal environmental/habitat Since the genes used in our study were selected for their conditions (Weber et al., 2003). Possibly, grass shrimp response to hypoxia, and not for their involvement in are following a similar strategy when exposed to chronic control of reproduction, the observed changes in gene hypoxia. expression do not provide insight into the molecular In various fish species, exposure to hypoxia results mechanisms through which hypoxia affects reproduction. in a decrease in gonadosomatic index (GSI), an indicator However, it is in combining the molecular biomarkers of reproductive readiness and future fecundity (Wu with whole animal responses such as fecundity, inter- et al., 2003; Thomas et al., 2006) as well as decreases in brood interval and embryo/larval survival that the data actual egg number (Landry et al., 2003). Furthermore, presented here become most valuable for understanding embryonic development, hatching success and survi- and predicting population-level effects of hypoxia. A vorship decreased with decreasing DO in Florida conceptual model for scaling molecular and reproductive flagfish, Jordanella floridae (Hale et al., 2003), and biomarkers of environmental stressors to the population there was a significant difference in hatching success level was presented by Brouwer et al. (2005).Thismodel between embryos exposed to normoxia and moderate stresses the importance of having both molecular as well (2–3 ppm DO) hypoxia. It appears grass shrimp em- as whole-animal inputs to be able to predict ecologically bryos are more tolerant of hypoxia than Florida flagfish relevant population effects. Thus, the data presented here embryos, as there was no difference in hatching success can be used in the development of physiological/sta- of grass shrimp in moderate hypoxia in either the single tistical, individual-based (IBM) and matrix projection brood or the multiple brood experiments. However, models (Rose et al., 2003) to gain a better understanding embryos of grass shrimp exposed to severe hypoxia took of population-level consequences of chronic hypoxia. longer to hatch than normoxic embryos, although there Chronic hypoxia of 1–3 d duration is not uncommon in was no difference in ultimate hatching success. This marsh systems where grass shrimp reside during summer contrasts with the Florida flagfish results, where no months along both the Gulf of Mexico and the South- embryos exposed to severe hypoxia (b1.0 ppm DO) eastern United States (see National Estuarine Research hatched (Hale et al., 2003). Thus, while chronic hypoxia Reserve water quality data, http://cdmo.baruch.sc.edu/ appears to have some adverse effects on grass shrimp data_summary.cfm), suggesting grass shrimp are exposed reproduction (longer interbrood interval in moderate to these conditions in their natural environments. Fur- hypoxia, longer hatch time in severe hypoxia), the thermore, a number of mobile, hypoxia tolerant estuarine higher fecundity, greater percentage of ovigerous fe- organisms have been shown to remain in hypoxic con- males and increased larval survival time of hypoxia- ditions (Pihl et al., 1991; Breitburg et al., 1994), sug- exposed grass shrimp indicates an overall strategy of gesting at least some species do not actively avoid such attempting to maximize reproduction in unfavorable conditions. Thus, our experimental results are applicable conditions. to natural marsh systems. Current research is focused on Grass shrimp appear to be quite tolerant of moderate cyclic hypoxia exposures in the laboratory and grass hypoxia, based on few significant changes in gene shrimp captured from hypoxic and normoxic field sites to expression and no differences in condition factor. Severe determine if the indicators determined from laboratory hypoxia appears to have short term effects, based on a experiments are useful in a field situation. Results from decrease in condition factor during the first 7 days of these studies will help to further refine development of exposure in combination with upregulation of genes im- models, and will continue to demonstrate the importance portant in oxygen transport and energy metabolism. In of combining molecular and whole animal data within the general, grass shrimp appear to adapt well to both same study. moderate and severe chronic hypoxia with little long-term mortality and few noticeable physiological effects. In- Acknowledgments deed, the reaction to hypoxia appears to be an increase in reproduction for this species, an unexpected result. We appreciate the technical expertise of B. Carter, T. However, this may be an artifact of the laboratory envi- Li, W. Grater, C. King and M. Peterson for their help 30 M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 with subtractive library construction, cDNA cloning, Brown-Peterson, N.J., Larkin, P., Denslow, N., King, C., Manning, sequencing, shrimp sampling and husbandry and C., Brouwer, M., 2005. Molecular indicators of hypoxia in the blue crab Callinectes sapidus. Mar. Ecol. Prog. Ser. 286, statistical consultation. This research was supported by 203–215. grants to M.B. from the US Environmental Protection Burnett, L.E., Burnett, K.G., 2000. The effects of hypoxia and hyper- Agency's Science to Achieve Results (STAR) Estuarine campnia on cellular defenses of oysters, shrimp and fish. Comp. and Great Lakes (EaGLe) program through funding to Biochem. Physiol., B 126, S20. the Consortium for Estuarine Ecoindicator Research for Buzzelli, C.P., Luettich, R.A., Powers, S.P.,Peterson, C.H., McNinch, J.E., Pinckney, J.L., Pearl, H.W., 2002. Estimating the spatial extent of the Gulf of Mexico (CEER-GOM; US EPA Agreement bottom-water hypoxia and habitat degradation in a shallow estuary. R82945801) and from the National Oceanic and Mar. Ecol. Prog. Ser. 230, 103–112. Atmospheric Administration (NAO3NOS4260216). Al- Choi, J., Roche, H., Caquet, T., 2000. Effects of physical (hypoxia, though the research described in this article was funded hyperoxia) and chemical (potassium dichromate, fenithrothion) wholly or in part by the US EPA, it has not been stress on antioxidant enzyme activities in Chironomus riparius Mg (Dipteria, Chironomidae) larvae: potential biomarkers. Environ. subjected to the Agency's required peer and policy Toxicol. Chem. 19, 495–500. review and, therefore, does not necessarily reflect the Coiro, L.L., Pucher, S.L., Miller, D.C., 2000. Hypoxic effects on views of the Agency and no official endorsement should growth of Palaemonetes vulgaris larvae and other species: using be inferred. [SS] constant exposure data to estimate cyclic exposure response. J. Exp. Mar. Biol. Ecol. 247, 243–255. Czyzyk-Krzeska, M.F., 1997. Molecular aspects of oxygen sensing in References physiological adaptation to hypoxia. Respir. Physiol. 110, 99–111. DeFur, P.L., Mangum, C.P., Reese, J.E., 1990. Respiratory responses Allison, D.B., Cui, X., Page, G.P., Sabripour, M., 2005. Microarray data of the blue crab Callinectes sapidus to long-term hypoxia. Biol. analysis: from disarray to consolidation and consensus. Nat. Rev., Bull. 178, 46–54. Genet. 7, 55–65. Diaz, R.J., Rosenberg, R., 1995. Marine benthic hypoxia: a review of Beckman, R.P., Mizzen, L.A., Welch, W.J., 1990. Interaction of Hsp 70 its ecological effects and the behavioural responses of benthic with newly-synthesized proteins: implications for protein folding macrofauna. Oceanogr. Mar. Biol. Annu. Rev. 33, 245–303. and assembly. Science 248, 850–854. Gracey, A.Y., Troll, J.V., Somero, G.N., 2001. Hypoxia-induced gene Bell, G.W., Eggleston, D.B., 2005. Species-specific avoidance response expression profiling in the euryoxic fish Gillichthys mirabilis. by blue crabs and fish to chronic and episodic hypoxia. Mar. Biol. Proc. Natl. Acad. Sci. U. S. A. 98, 1993–1998. 146, 761–770. Hale, R.E., St. Mary, C.M., Lindstrom, K., 2003. Parental responses to Bell, G.W., Eggleston, D.B., Wolcott, T.G., 2003a. Behavioral responses changes in costs and benefits along an environmental gradient. of free-ranging blue crabs to episodic hypoxia. I. Movement. Mar. Environ. Biol. Fisches 67, 107–116. Ecol. Prog. Ser. 259, 215–225. Hochachka, P.W., Lutz, P.L., 2001. Mechanism, origin and evolution Bell, G.W., Eggleston, D.B., Wolcott, T.G., 2003b. Behavioral of anoxia tolerance in animals. Comp. Biochem. Physiol., B 130, responses of free-ranging blue crabs to episodic hypoxia. II. 435–459. Feeding. Mar. Ecol. Prog. Ser. 259, 227–235. Hochachka, P.W., Buck, L.T., Doll, C.J., Land, S.C., 1996. Unifying Benjamin, I.J., Kroger, B., Williams, S., 1990. Activiation of the heat theory of hypoxia tolerance; molecular/metabolic defense and rescue shock transcription factor by hypoxia in mammalian cells. Proc. mechanisms for surviving oxygen lack. Proc. Natl. Acad. Sci. U. S. A. Natl. Acad. Sci. U. S. A. 87, 6263–6267. 93, 9493–9498. Boleza, K.A., Burnett, L.E., Burnett, K.G., 2001. Hypercapnic hypoxia Karlin, S., Altschul, S.F., 1990. Methods for assessing the statistical compromises bactericidal activity of fish anterior kidney cells significance of molecular sequence features by using general against opportunistic environmental pathogens. Fish Shellfish scoring schemes. Proc. Natl. Acad. Sci. U. S. A. 87, 2264–2268. Immunol. 11, 593–601. Kothapalli, R., Yoder, S.J., Mane, S., Loughran, T.P., 2002. Microarray Breitburg, D.L, Steinberg, N.D., DuBeau, S., Cooksey, C., Houde, E.D., results: how accurate are they? BMC Bioinformatics 3, 22. 1994. Effects of low dissolved oxygen on predation on estuarine fish Kultz, D., 2003. Evolution of the cellular stress proteome: from mono- larvae. Mar. Ecol. Prog. Ser. 104, 235–246. phyletic origin to ubiquitous function. J. Exp. Biol. 206, 3119–3124. Brouwer, M., Hoexum Brouwer, T., Grater, W., Enghild, J., Landry, C.A., Manning, S., Cheek, A.O., 2003. Hypoxia affects Thogersen, I., 1997. The paradigm that all oxygen-respiring reproduction in Gulf killifish (Fundulus grandis). Integr. Comp. eukaryotes have cytosolic CuZn-superoxide dismutase and that Biol. 43, 812. Mn-superoxide dismutase is localized to the mitochondria does Larkin, P., Folmar, L.C., Hemmer, M.J., Poston, A.J., Denslow, N.D., not apply to a large group of marine arthropods. Biochem. 36, 2003. Expression profiling of estrogenic compounds using a 13381–13388. sheepshead minnow cDNA macroarry. Environ. Health Perspect. Brouwer, M., Hoexum Brouwer, T., Grater, W., Brown-Peterson, N., Toxicogenomics 111, 839–846. 2003. A novel cytosolic Mn-superoxide dismutase (MnSOD) has Luo, T., Zhang, X., Shao, Z., Xu, X., 2003. PmAV, a novel gene replaced cytosolic Cu, ZnSOD in crustacea that use copper (hemo- involved in virus resistance of shrimp Penaeus monodon. FEBS cyanin) for oxygen transport. Biochem. J. 374, 219–228. Lett. 551, 53–57. Brouwer, M., Brown-Peterson, N.J., Larki, P., Manning, S., Denslow, N., Mangum, C.P., 1997. Adaptation of the oxygen transport system to Rose, K., 2005. Molecular and organismal indicators of chronic and hypoxia in the blue crab, Callinectes sapidus. Am. Zool. 37, 604–611. intermittent hypoxia in marine crustacea. In: Bortone, S.A. (Ed.), Manning, C.S., Schesny, A.L., Hawkins, W.E., Barnes, D.H., Barnes, Estuarine Indicators. CRC Press, Boca Raton, FL, pp. 261–276. D.S., Walker, W.W., 1999. Exposure methodologies and systems M. Brouwer et al. / Journal of Experimental Marine Biology and Ecology 341 (2007) 16–31 31

for long term chemical carcinogenicity studies with small fish Ton, C., Stamatiou, D., Dzau, V.J., Liew, C.-C., 2002. Construction of species. Toxicol. Methods 9 (3), 201–217. a zebrafish cDNA microarray: gene expression profiling of the McMahon, B.R., 2001. Respiratory and circulatory compensation to zebrafish during development. Biochem. Biophys. Res. Commun. hypoxia in crustaceans. Respir. Physiol. 128, 349–364. 296, 1134–1142. Meehean, O.L., 1936. Notes on the freshwater shrimp Palaemonetes Ton, C., Stamatiou, D., Liew, C.-C., 2003. Gene expression profile of paludosa (Gibbes). Trans. Am. Microsc. Soc. 55, 433–441. zebrafish exposed to hypoxia during development. Physiol. Geno- Mistri, M., 2004. Effects of hypoxia on predator-prey interactions mics 13, 97–106. between juvenile Carcinus aestuarii and Musculista senhousia. van der Meer, D.L.M., van den Thillart, G.E.E.J.M., Witte, F., de Mar. Ecol. Prog. Ser. 275, 211–217. Bakker, M.A.G., Besser, J., Richardson, M.K., Spaink, H.P., Leito, Murphy, B.R., Willis, D.W. (Eds.), 1996. Fisheries Techniques, 2nd ed. J.T.D., Batowski, C.P., 2005. Gene expresión profiling of the long- American Fisheries Society, Bethesda, Maryland. term adaptive response to hypoxia in the gills of adult zebrafish. Napier, I., Ponka, P., Richardson, D.R., 2005. Iron trafficking in the Am. J. Physiol., Regul. Integr. Comp. Physiol. 289, 1512–1519. mitochondrion: novel pathways revealed by disease. Blood 105, Vijayasarathy, C., Damle, S., Prabul, S.K., Otto, C.M., Avadhani, N.G., 1867–1874. 2003. Adaptive changes in the expression of nuclear and mito- Paul, R.J., Zeis, B., Lamkemeyer, T., Seidl, M., Pirow, R., 2004. chondrial encoded subunits of cytochrome c oxidase and the Control of oxygen transport in the microcrustacean Daphnia: catalytic activity during hypoxia. Eur. J. Biochem. 270, 871–879. regulation of haemoblobin expression as central mechanism of Wannamaker, C.M., Rice, J.A., 2000. Effects of hypoxia on move- adaptation to different oxygen and temperature conditions. Acta ments and behavior of selected estuarine organisms from the Physiol. Scand. 182, 259–275. southeastern United States. J. Exp. Mar. Biol. Ecol. 249, 145–163. Pihl, L., Baden, S.P., Diaz, R.J., 1991. Effects of periodic hypoxia on Weber, A., Vesela, S., Repka, S., 2003. The supposed lack of trade-off distribution of demersal fish and crustaceans. Mar. Biol. 108, among Daphnia galeata life history traits is explained by increased 349–360. adult mortality in Chaoborus conditioned treatments: Recent Rose, K.A., Murphy, C.A., Diamond, S.L., Fuiman, L.A., Thomas, P., Developments in Fundamental and Applied Plankton Research 2003. Using nested models and laboratory data for predicting (Guest Editors: R. Van Donk, M. Boersman and P. Spaak). Hydro- population effects of contaminants on fish: a step toward a bottom- biologia 491, 273–287. up approach for establishing causality in field studies. Hum. Ecol. Welsh, B.L., 1975. The role of grass shrimp, Palaemonetes pugio,ina Risk Assess. 9, 231–257. tidal marsh ecosystem. Ecology 56, 513–530. Russel, W.J., Ho, Y.S., Parish, G., Jackson, R.M., 1995. Effects of Williams, R.S., 1986. Mitochondrial gene expression in mammalian hypoxia on MnSOD expression in mouse lungs. Am. J. Physiol., striated muscle. Evidence that variation in gene dosage is the major Lung Cell. Mol. Physiol. 13, L221–L225. regulatory event. J. Biol. Chem. 261, 12390–12394. Sokal, R.R., Rohlf, F.J., 1995. Biometry, 3rd ed. W.H. Freeman, New York. Wong, M.L., Medrano, J.F., 2005. Real-time PCR for mRNA Tankersley, R.A., Wieber, M.G., 2000. Physiological responses of quantitation. BioTechniques 39, 1–11. postlarval and juvenile blue crabs Callinetes sapidus to hypoxia Wu, R.S.S., 2002. Hypoxia: from molecular responses to ecosystem and anoxia. Mar. Ecol. Prog. Ser. 194, 179–191. responses. Mar. Pollut. Bull. 45, 35–45. Taketani, S., 2005. Acquisition, mobilization and utilization of cellular Wu, R.S.S., Lam, P.K.S., Wan, K.L., 2002. Tolerance to, and avoidance iron and heme: endless findings and growing evidence of tight of, hypoxia by the penaeid shrimp (Metapenaeus ensis). Environ. regulation. Tohoku J. Exp. Med. 205, 297–318. Pollut. 118, 351–355. Terwilliger, N.B., 1998. Functional adaptations of oxygen-transport Wu, R.S.S., Zhou, D.S., Randall, D.L., Woo, N.Y.S., Lam, P.K.S., proteins. J. Exp. Biol. 201, 1085–1098. 2003. Aquatic hypoxia is an endocrine disruptor and impairs fish Thomas, P., Rahman, S.Md., Kummer, J.A., Lawson, S., 2006. reproduction. Environ. Sci. Technol. 37, 1137–1141. Reproductive endocrine dysfunction in Atlantic croaker exposed to Zarate, J., Bradley, T.M., 2003. Heat shock proteins are not sensitive hypoxia. Mar. Environ. Res. 62, S249–S252. indicators of hatchery stress in salmon. Aquaculture 223, 175–187.